The QuantMinds Hackathon:
AI Safety and Security
Join your peers at the third annual QuantMinds hackathon in partnership with CompatibL and get your chance to prove your skills as you code to compete with the top quant minds in the industry.
Monday 16 November 2026
Apply your skills to real world problems
No coding skills required.
Who's it for?
Open to all industry professionals, academics, and students.
Team or individual entries are accepted (one entry per individual).
Prompts only, no coding required.
Why should I join?
The hackathon is a chance to demonstrate and hone your skills and learn new modelling techniques.
Your solution may have a tangible impact on the financial industry and could become a hot topic at the next QuantMinds conference.
What's the prize?
Awards will be announced for each of the four categories, with a second set of prizes for the best student participant or team in each category (post-graduate programs included).
The overall grand prize and overall student grand prize will be awarded based on the top score relative to the category average.
How to prepare?
Participants will use their own Anthropic API key during the competition. Getting the key in advance is highly recommended.
Contact Ella at ella.burton@informa.com to apply.
Everything you need to know...
Each hackathon stream is an AI safety and security exercise where the red team plays the attacker, trying to get AI agents to provide incorrect responses, and the blue team plays the defender, working to stop these attacks.
Both the red and blue teams will influence the outcome through prompts only (no coding required).
The organisers will be the red team. You will be the blue team.
- Your goal is to get the model to resist the red team's efforts to elicit an incorrect AI agent response
- Each of the four streams will have a task, dataset, and a list of red team prompts
- Half of the red team prompts will be revealed at the start of the hackathon and used for development, and the other half will be embargoed and used for scoring
- Your prompt will be provided to the AI agent first, followed by the task and then the red team’s prompt
- Your goal is to construct the prompt such that the AI agent produces the correct response despite the red team’s prompt
- Your score is the fraction of runs where your prompt restores the correct AI agent response despite the red team’s prompt
Hackathon Leader Alexander Sokol
Alexander Sokol is the founder, Executive Chairman, and Head of Quant Research at CompatibL, a trading and risk technology firm. He previously co-founded Numerix, serving as CTO from 1996 to 2003.He received the Quant of the Year Award in 2018 alongside Leif Andersen and Michael Pykhtin for their work on settlement gap risk, even in the presence of initial margin. His influential research includes over 50 publications in theoretical physics and quantitative finance, including:
- Systemic wrong-way risk (with Pykhtin)
- Joint measure models and local price of risk (with John Hull and Alan White)
- Mean reversion skew (Risk Books)
Sokol entered college at 14 and earned a Ph.D. in Theoretical Physics by 22. He won the USSR Academy of Sciences Medal for Best Student Research in 1988.
Notes
*While there is no requirement to publish the competition entries, confidential treatment of code and prompts is not practical in a competition setting. For this reason, eligible competition entries must be submitted under an open-source license.